86+ AI Cheating Statistics 2026: Academic Misconduct Rates & Trends

Universities spent years perfecting plagiarism detection. Then generative AI arrived, and the entire enforcement model broke.

AI cheating statistics from 2025 show nearly 7,000 UK university students were formally caught cheating with AI tools in the 2023-24 academic year alone, up from 1.6 cases per 1,000 students to 5.1, a more than threefold increase in a single year. That figure counts only the students who got caught.

Here is what the data actually shows about how widespread this is, which institutions are feeling it most, and how far detection has (and has not) come.

Key AI Cheating Statistics for 2025

AI cheating has moved from a fringe concern to a documented crisis: formal misconduct cases in UK universities more than tripled in a single academic year, and the pattern is spreading globally.

• Nearly 7,000 UK university students were formally caught using AI to cheat in 2023-24, equivalent to 5.1 cases per 1,000 students (up from 1.6 per 1,000 the prior year)
• AI cheating incidents rose from 1.6 per 1,000 students in 2022-23 to 7.5 per 1,000 in 2024-25, according to Anara’s 2025 higher education report
• 86% of students globally use AI tools in their studies, per the Digital Education Council Global AI Student Survey 2024 of 3,839 students across 16 countries
• 18% of UK undergraduate students admit to submitting AI-generated text in their assignments, per 2025 HEPI research
• 95% of the academic community believes AI is being misused at their institutions, per a 2025 study by Turnitin and Vanson Bourne
• 1 in 10 writing assignments reviewed by Turnitin’s AI detection tool showed some evidence of AI use, with 3 in 100 generated mostly by AI
• 68% of educators now use AI detection tools, up substantially from the previous year, per the Center for Democracy and Technology
• 35% of UK students have used AI in a school learning context, the highest rate across surveyed European countries, per the Future of Education Report 2025

Student AI Usage Patterns and Statistics

In one year, the share of UK undergraduates using AI for assessments jumped from 53% to 88%. That is not gradual adoption; that is a near-complete shift in how a generation approaches academic work. The 64% of students now using AI specifically to generate text (up from 30% in 2024) is where this tips from tool use into academic integrity territory.

Metric	2024	2025	Source
Students using AI for assessments	53%	88%	HEPI / Kortext Survey 2025
Undergraduates using AI for academic work (any form)	66%	92%	HEPI / Kortext Survey 2025
Students using AI to generate text	30%	64%	HEPI / Kortext Survey 2025
Students submitting AI-generated text directly	Not reported	18%	HEPI / Kortext Survey 2025
U.S. college students using gen AI chatbot at least weekly	Not reported	65%	Time for Class Survey, Spring 2025

Text generation doubled as a use case in a single academic year. But not every AI interaction is a misconduct incident. The student AI cheating statistics that matter most are the ones that show what students are doing with the output. Explaining concepts is the most common individual use case among UK undergraduates, per HEPI, which places much of this activity closer to tutoring than to cheating. The breakdown of use cases shows where the line sits:

• 58% of UK undergraduates use AI to explain concepts, the single most common use case, up from 36% in 2024 (HEPI / Kortext Survey 2025)
• 64% use AI to generate text, more than doubling from 30% in 2024 (HEPI / Kortext Survey 2025)
• 18% admit to submitting AI-generated text directly in assignments (HEPI / Kortext Survey 2025)
• 30% of U.S. college students used ChatGPT specifically for schoolwork in the 2022-23 academic year (Intelligent.com survey, 1,223 students)

The 40-point gap between students generating text (64%) and students submitting it directly (18%) suggests most are using AI as a drafting aid rather than a wholesale replacement. Whether institutions treat that distinction as meaningful is the question reshaping academic policy right now.

AI Detection and Enforcement Statistics

Turnitin claims 98% accuracy for its AI detection tool, based on its own internal testing, with a false positive rate of under 1%. Those figures are self-reported, and the real-world record tells a different story. Since launching in April 2023, the tool has reviewed over 200 million papers, and 17% of all submissions between January and August 2024 showed more than 20% likely AI-generated content, up from 11% in the tool’s first year. The scale of what detection tools are being asked to process has grown faster than confidence in their reliability.

Metric	Value	Source / Period
Turnitin claimed detection accuracy	98%	Turnitin internal testing (2025)
Turnitin claimed false positive rate	Less than 1%	Turnitin internal testing (2025)
GPTZero false positive rate (standard testing)	1–2%	Standard testing scenarios
Total papers reviewed by Turnitin AI detector	Over 200 million	Since April 2023 (Campus Technology)
Papers flagged with more than 20% AI content (2024)	17% of submissions	Turnitin, Jan–Aug 2024
Papers flagged with more than 80% AI content (2024)	5% of submissions	Turnitin 2024 Wrapped report
Teachers using AI detection tools (2023–24)	68%	Center for Democracy and Technology

The 68% of middle and high school teachers using AI detection tools in 2023–24 (up from 38% the prior year) reflects a system moving fast to respond. But some of the largest adopters have since reversed course, citing false accusation rates that damaged trust without improving integrity. The AI cheating detection statistics that matter most right now are not about accuracy claims; they are about institutional confidence:

• Australian Catholic University reported nearly 6,000 AI cheating allegations in 2024, roughly 90% of all academic integrity cases; around 25% of all referrals were dismissed after investigation, leading the university to abandon Turnitin’s AI detection tool in March 2025
• The University of Cape Town announced in July 2025 it would stop using Turnitin’s AI Score from October 1, 2025, citing evidence that the tools are unreliable and not fit for purpose

When a tool flags 25% of its own cases as wrongful accusations at scale, the enforcement problem does not get solved by better software. It gets handed back to faculty with no clearer standard than before.

AI Cheating Consequences and Discipline Statistics

Formal penalties for AI cheating are no longer rare outcomes; they are becoming the default. 5.1 per 1,000 UK university students were formally caught cheating with AI in 2023-24 (up from 1.6 per 1,000 the prior year), according to Freedom of Information data obtained by The Guardian from 131 universities. At the same time, 63% of middle and high school teachers reported that students had gotten in trouble for AI use accusations during 2023-24, up from 48% the year before, per the Center for Democracy and Technology.

Institution / Cohort	2022-23 Cases	2023-24 Cases	Penalties Issued
UK universities overall (per 1,000 students)	1.6 per 1,000	5.1 per 1,000	Not disaggregated
Queen Mary University of London	10 suspected, 9 penalties	89 suspected	89 (100% conversion)
University of Sheffield	6 cases	92 cases	79 penalties issued
US middle and high schools (teachers reporting student discipline)	48% of teachers	63% of teachers	Center for Democracy and Technology

The Queen Mary figure is the sharpest illustration of where enforcement is heading. Every single suspected case in 2023-24 resulted in a penalty, a conversion rate that reflects institutions moving away from case-by-case discretion toward near-automatic consequences. The University of Sheffield’s fifteenfold case increase in one year, with 79 of 92 suspected students penalized, points in the same direction. These AI cheating consequences and discipline statistics come from Times Higher Education Freedom of Information data covering all 24 Russell Group universities, which means the pattern holds across the UK’s most research-intensive institutions, not just outliers.

AI Cheating by Student Demographics: Usage Patterns and Statistics

Teen use of ChatGPT for schoolwork doubled in a single year, rising from 13% in 2023 to 26% in 2024. That acceleration did not hit all students equally. Grade level, gender, field of study, and household income each predict whether a student reaches for an AI tool, and by how much.

The grade-level gradient is the clearest pattern in the data. Among U.S. secondary students, ChatGPT use for school climbs steadily with age: 20% for 7th and 8th graders, 26% for 9th and 10th graders, and 31% for 11th and 12th graders, according to Pew Research Center data. By the time students reach college, 84% of U.S. high schoolers reported using generative AI tools for schoolwork as of May 2025, up from 79% in January, per College Board surveys conducted across the 2024–2025 academic year.

• Teen use of ChatGPT for schoolwork rose from 13% in 2023 to 26% in 2024, a full doubling in one academic year (Pew Research Center)
• 84% of U.S. high school students used generative AI for schoolwork by May 2025, up from 79% in January 2025 (College Board)
• 69% of U.S. high school students specifically used ChatGPT for assignments and homework (College Board, June 2024–June 2025)
• 45% of all ChatGPT users globally are under 25 years old
• Male students, STEM and health course students, and more socioeconomically advantaged students are significantly more likely to use generative AI than their peers (HEPI / Kortext Student Generative AI Survey 2025, 1,041 UK undergraduates)
• Women express stronger concern than men about the risk of AI misconduct accusations and about biased or false AI outputs (HEPI / Kortext Survey 2025)

AI Tool	Student Usage Share	Source
ChatGPT	66%	Digital Education Council Global AI Student Survey 2024
Grammarly	25%	Digital Education Council Global AI Student Survey 2024
Microsoft Copilot	25%	Digital Education Council Global AI Student Survey 2024
Average number of AI tools per student	~2 tools	Digital Education Council Global AI Student Survey 2024
Students using any AI tool daily	25%	Digital Education Council Global AI Student Survey 2024
Students using any AI tool weekly	54%	Digital Education Council Global AI Student Survey 2024

The gender gap in AI tool adoption has shifted more than most reporting reflects. In January 2024, approximately 63% of ChatGPT users (among those with classifiable names) were male, per NBER analysis. By July 2025, OpenAI’s own published research showed female users had risen to 52% of the user base: a full inversion in roughly 18 months. The HEPI data suggests the underlying cause is not that women became more comfortable with AI broadly, but that their specific concerns (about false accusations and unreliable outputs) were never really about the tools themselves. They were about how institutions would respond when things went wrong.

AI Detection Tool Accuracy and Limitations Statistics

AI detection tools are being adopted at scale by institutions that believe the accuracy claims on the label. Those claims do not survive contact with two real-world conditions: paraphrasing and non-native English writing. A 2023 NeurIPS study found that a paraphrasing tool called DIPPER dropped DetectGPT’s detection accuracy from 70.3% to just 4.6% at a constant false positive rate of 1%. That is not a marginal performance dip; it is near-total failure triggered by a freely available text-spinning tool any student can use.

Tool / Condition	Claimed or Measured Accuracy	False Positive Rate	Source
Turnitin (self-reported, internal testing)	Up to 98%	Less than 1%	Turnitin / BestColleges, 2025
GPTZero (controlled testing)	Not specified	1–2%	Standard controlled testing
DetectGPT (standard conditions)	70.3%	1%	NeurIPS 2023
DetectGPT after DIPPER paraphrasing attack	4.6%	1%	NeurIPS 2023
Seven AI detectors (TOEFL essays, non-native speakers)	61.22% falsely flagged as AI	61.22%	Stanford / journal Patterns, 2023
Seven AI detectors (TOEFL essays, 2026 replication)	Improved, but gap persists	23.1%	ACL Anthology, 2026

The bias problem cuts along the same lines as the paraphrasing problem: neither requires a student to do anything unusual to trigger a false result. The AI detection tool accuracy statistics from the Stanford study reveal just how uneven the risk is for non-native English writers specifically:

• A Stanford University study in the journal Patterns found that seven popular AI detectors classified 61.22% of TOEFL essays written by non-native English students as AI-generated
• All seven detectors unanimously flagged 19% of those non-native English essays as AI-generated, compared to a mean false positive rate of just 5.1% for essays written by U.S.-born students
• At least one detector flagged 97% of the non-native English essays as AI-generated
• A 2026 replication study published in ACL Anthology found the false positive rate for non-native English writing had fallen from 61.3% to 23.1%, improved but still substantially higher than the rate for native English speakers
• The AI detection tool market is projected to grow from $359.8 million (2020) to $1.02 billion (2028), even as institutional confidence in the tools is visibly eroding

The Reality Gap in AI Detection Accuracy

Independent research consistently finds that AI detectors perform at a fraction of the accuracy their vendors claim. A 2024 study published in the International Journal of Educational Technology in Higher Education (Springer Nature) ran 805 tests across six adversarial techniques and found the average detection accuracy on unmodified AI content was only 39.5%. Apply a simple manipulation, and that figure drops to 22.14%. The authors’ conclusion was direct: these tools cannot currently be recommended for determining academic integrity violations.

Study	Condition	Detection Accuracy
Springer Nature / Int’l Journal of Educational Technology in Higher Education (2024), 805 tests	Non-manipulated AI content	39.5%
Springer Nature / Int’l Journal of Educational Technology in Higher Education (2024), 805 tests	Adversarial techniques applied	22.14%
Perkins et al. (2024), Research Square	Standard conditions	Baseline (used as control)
Perkins et al. (2024), Research Square	Paraphrasing and synonym replacement applied	17.4 percentage points lower than baseline
Perkins et al. (2024), Research Square	Most effective adversarial techniques	12–15% accuracy

Two independent research teams, different methodologies, same direction: accuracy does not degrade gradually under realistic conditions; it collapses. The Perkins et al. finding that some techniques reduce detection to 12–15% is significant because synonym replacement and paraphrasing require no technical skill. A student with access to any free rewriting tool can reduce a detector’s reliability to near-random performance. That is the gap between a vendor’s internal accuracy figure and what the peer-reviewed AI cheating detection statistics actually show.

Why AI Detection False Positives Matter at Scale

A 1% false positive rate sounds like a quality control footnote. Applied to the volume of student writing that flows through AI detection tools each year, it becomes a wrongful accusation engine. U.S. first-year college students write an estimated 22.35 million essays annually. At a 1% false positive rate, that is 223,500 essays flagged as AI-generated that were written entirely by humans, according to NIU’s Center for Innovative Teaching and Learning (December 2024). The student on the receiving end of one of those flags does not experience a percentage. They experience a misconduct charge.

• A Bloomberg test of GPTZero and CopyLeaks on 500 pre-generative-AI essays found false positive rates of 1–2%, and likely higher, per NIU’s Center for Innovative Teaching and Learning (December 2024)
• At 1% across 22.35 million first-year college essays written annually in the U.S., approximately 223,500 essays would be falsely flagged as AI-generated each year
• Consequences for falsely flagged students include academic penalties, loss of scholarships, and lasting damage to academic records and future opportunities

False Positive Rate Scenario	Essays Falsely Flagged per Year (U.S. First-Year Students)	Basis
0.5% (Turnitin’s claimed rate, halved)	Approximately 111,750	22.35M essay estimate, NIU CITL (December 2024)
1% (Bloomberg test lower bound)	Approximately 223,500	Bloomberg / NIU CITL (December 2024)
2% (Bloomberg test upper bound)	Approximately 447,000	Bloomberg / NIU CITL (December 2024)
In a university of 20,000 students at 1–2%	200–400 students per semester	NIU CITL illustrative scale (December 2024)

Institutional Response and Discipline Rate Statistics

Only 28% of teachers have received guidance on how to respond when they suspect a student has used generative AI inappropriately, according to the Center for Democracy and Technology’s “Up in the Air” report, as cited in a December 2024 U.S. Commission on Civil Rights report on AI in K-12 education. That figure sits alongside a discipline rate that has climbed sharply. The gap between how often students are being penalized and how often educators feel equipped to make that call is the central tension in institutional AI governance right now.

Policy Status	Share of Institutions	Source
Formal AI policy already in force	19%	UNESCO global survey, Sept. 2025 (400 institutions, 90 countries)
Guiding AI framework under development	42%	UNESCO global survey, Sept. 2025
Have or are developing formal AI guidance (combined)	66%	UNESCO global survey, Sept. 2025
No formal guidance yet	34%	UNESCO global survey, Sept. 2025

The UNESCO data covers 400 institutions across 90 countries. Two-thirds have a policy or are actively building one. But a policy document and a trained faculty body are different things. The institutional response and discipline rate statistics from RAND’s American School District Panel reveal a starker version of the same problem at the K-12 level, and the gap runs along income lines:

• 67% of low-poverty U.S. school districts had provided teacher training on AI use by fall 2024 (RAND Corporation, 2024–2025 school year)
• 42% of middle-poverty districts had provided the same training by fall 2024 (RAND Corporation)
• 39% of high-poverty districts had provided teacher AI training by fall 2024 (RAND Corporation)
• 28% of teachers overall have received any guidance on how to respond to suspected AI misuse, per the Center for Democracy and Technology’s “Up in the Air” report

The training gap by district income means the institutions least resourced to handle AI misconduct are also the ones most likely to handle it inconsistently. A student in a high-poverty district faces a teacher with no formal guidance and no training. A student in a low-poverty district is more likely to face one who has both. That disparity does not show up in misconduct statistics, but it shapes them.

Global AI Cheating Trends and Regional Differences Statistics

The UK leads the world in documented AI cheating cases. It almost certainly does not lead the world in actual AI cheating. Almost 7,000 proven AI misconduct cases were tracked across 131 UK universities in the 2023-24 academic year, at a rate of 5.1 cases per 1,000 students, up from 1.6 per 1,000 the prior year, according to a Guardian investigation using Freedom of Information Act requests. That tripling happened inside the world’s most systematically monitored higher education system. What it reveals about the UK is less about student behavior and more about institutional capacity to catch it.

Region / Country	Key Documented Metric	Policy Orientation	Source
United Kingdom	~7,000 proven cases across 131 universities in 2023-24; 5.1 per 1,000 students (up from 1.6 per 1,000)	Academic integrity and originality enforcement	The Guardian / FOI, June 2025
United States	26% of teens used ChatGPT for schoolwork in 2024 (up from 13% in 2023); no national case tracking system	Leveraging AI to enhance teaching and learning	Pew Research Center, Jan. 2025; Jin et al. (2024) via Turnitin
Australia	At least a dozen universities using AI detection software; documented errors costing students grades and graduation standing	Academic integrity enforcement (similar to UK)	ABC News, Oct. 2025
Hong Kong	Policy studies in 40-university cross-regional analysis	Leveraging AI to enhance teaching and learning	Jin et al. (2024) via Turnitin
Global (16 countries)	86% of students use AI tools in their studies; 54% use AI weekly; nearly 1 in 4 use it daily	Varies by institution	Digital Education Council Global AI Student Survey 2024 (3,839 students)

The 86% global AI usage figure cuts across every regional policy distinction. Students in countries with strict enforcement frameworks and students in countries with permissive or ambiguous ones are using these tools at comparable rates. What diverges is not the behavior but the institutional response to it. A study of 40 universities across six global regions found that UK and Australian institutions frame the issue primarily as an academic integrity and originality problem, while US and Hong Kong institutions lean toward AI as a teaching enhancement tool. That framing difference shapes which cases get counted and which disappear into the gap between policy and practice.

Australia’s situation adds a specific complication. At least a dozen universities are actively deploying AI detection software, but ABC News reporting from October 2025 confirmed that documented errors from those tools have already cost students grades, caused failed subjects, and threatened graduation timelines at institutions including Queensland University of Technology and the University of Melbourne. Deploying detection infrastructure does not resolve the regional data gap. It adds a new one: cases where the tool was wrong.

Impact of AI Cheating on Institutions: Costs and Consequences

Sixty percent of higher education leaders say cheating has increased since generative AI became widely available. Most of them are also paying to address it without being sure they can even see it clearly: 54% say their faculty are not effective at recognizing AI-generated content, according to a survey of higher education executives conducted by AAC&U and Elon University between November and December 2024. That combination, rising misconduct and unreliable detection capacity, is what has turned an academic integrity problem into a budget problem.

• Each misconduct investigation costs institutions an average of $3,200 to $8,500 when administrative time, legal reviews, and academic committee proceedings are factored in
• A single large university handling 200 cases annually faces investigation costs alone reaching $1.7 million
• Some colleges spend upward of $50,000 annually on faculty training programs to help educators identify AI-generated work
• Policy overhauls cost institutions between $15,000 and $75,000 per comprehensive AI-specific academic integrity policy, covering consultant and legal expert fees
• California Community Colleges lost $11 million in financial aid funds to applicant fraud in 2024, with 31% of applications identified as fraudulent (BCG analysis, EdSource, September 2025)
• Institutions experiencing publicized academic integrity breaches have seen average enrollment drops of 8–12% over the following two years

Cost Category	Estimated Cost Range	Context
Per-case misconduct investigation	$3,200 – $8,500	Includes administrative, legal, and committee costs
Annual investigation costs (200-case university)	Up to $1.7 million	Investigation costs only, before detection or prevention
Annual faculty AI training (per institution)	Upward of $50,000	Training to identify AI-generated academic work
AI-specific policy overhaul (per policy)	$15,000 – $75,000	Consultant and legal expert fees
Total AI misconduct budget (mid-sized university)	3 – 7% of annual operating budget	Detection, investigation, and prevention combined
For a university with a $100M operating budget	Up to $7 million annually	Based on 7% allocation figure

The AI cheating impact statistics from California add a dimension that goes beyond campus misconduct systems. When fraudulent applications at a community college system reach 31%, the financial aid pipeline itself becomes compromised. AI-based fraud detection tools identified twice as many fraudulent applications as manual review in that same system, per BCG analysis, which points toward where institutional spending is likely to shift next: not just toward catching students who cheat on assignments, but toward screening who enters the system in the first place.

Future Trends and Projections in AI Cheating Statistics

Institution-wide AI adoption in higher education jumped from 49% to 66% in a single year, and 88% of higher education institution respondents expect their institution’s AI use to keep rising over the next two years, according to Ellucian’s 2025 AI in Higher Education survey. That number is not a projection built on assumptions. It reflects what the people running these institutions see happening inside them right now. The question is no longer whether AI becomes standard in academic life. It is whether the support structures around it catch up before the integrity gaps widen further.

Assessment frameworks are already shifting in response. In the UK, 59% of undergraduates said the way they are assessed has changed “a lot” because of generative AI, per the HEPI and Kortext Student Generative AI Survey 2025 (1,041 full-time UK undergraduates). The proportion of students saying university staff are well-equipped to work with AI doubled in twelve months, from 18% in 2024 to 42% in 2025. That is meaningful progress. It also means 58% of students still do not believe their institutions are ready.

Trend Metric	Earlier Measure	Current / Projected Measure	Source
Institution-wide AI adoption (HE sector)	49% (2024)	66% (2025)	Ellucian 2025 AI in Higher Education survey
HEI respondents expecting AI use to keep rising	Not reported	88% over next two years	Ellucian 2025 AI in Higher Education survey
Students saying staff are well-equipped for AI	18% (2024)	42% (2025)	HEPI / Kortext Student Generative AI Survey 2025 (1,041 UK undergraduates)
UK undergraduates who say assessment has changed “a lot” due to AI	Not reported	59% (2025)	HEPI / Kortext Student Generative AI Survey 2025
U.S. teens using AI chatbots for schoolwork (Pew 2026)	26% used ChatGPT for schoolwork (2024)	54% use AI chatbots for schoolwork (early 2026)	Pew Research Center, 1,458 U.S. teens ages 13–17
Teens reporting peers use AI to cheat “often”	Not reported	60% (early 2026)	Pew Research Center, 1,458 U.S. teens ages 13–17

The Pew 2026 data on U.S. teens is the sharpest early indicator of where the next phase lands. Among 13-to-17-year-olds, 54% already use AI chatbots for schoolwork, and 10% use AI for all or most of their homework. That cohort is two to five years from higher education. The academic integrity frameworks being built now will be tested by students for whom AI assistance was never experimental. It was just how school worked.

Students themselves are not uniformly comfortable with that trajectory. The concern data from HEPI 2025 and Turnitin’s analysis of AI-native academic integrity trends shows a generation that is using these tools while remaining genuinely uncertain about what they are doing to their own development:

• 59% of students worry that AI could reduce their critical thinking skills (HEPI 2025 / Turnitin analysis)
• 49% are concerned about becoming too dependent on AI tools (HEPI 2025 / Turnitin analysis)
• 35% of students report receiving any institutional support to develop AI skills, leaving the majority navigating these tools without formal guidance (HEPI 2025 / Turnitin analysis)
• 10% of U.S. teens report using AI for all or most of their homework (Pew Research Center, 1,458 teens ages 13–17, early 2026)
• 60% of U.S. teens say students at their schools use AI to cheat often (Pew Research Center, early 2026)

Sources

• https://www.theguardian.com/education/2025/jun/15/thousands-of-uk-university-students-caught-cheating-using-ai-artificial-intelligence-survey (2025-06-15)
• https://cdn.prod.website-files.com/65f1d299b87bcc50550a6398/66a46f7b2a6137a25d3f69bb_Digital%20Education%20Council%20Global%20AI%20Student%20Survey%202024.pdf (2024-08-01)
• https://www.hepi.ac.uk/reports/student-generative-ai-survey-2025/ (2025-02-26)
• https://edscoop.com/a-third-of-college-students-using-chatgpt-on-homework-survey-finds/ (2023-05-26)
• https://www.universityworldnews.com/post.php?story=20250228175423255 (2025-02-28)
• https://www.insidehighered.com/news/student-success/academic-life/2025/06/11/65-percent-students-use-gen-ai-chat-bot-weekly (2025-06-11)
• https://www.edweek.org/technology/new-data-reveal-how-many-students-are-using-ai-to-cheat/2024/04 (2024-04-05)
• https://www.bestcolleges.com/news/analysis/testing-turnitin-new-ai-detector/ (2025-07-08)
• https://www.turnitin.com/blog/2024-turnitin-wrapped (2024-12-31)
• https://www.abc.net.au/news/2025-10-09/artificial-intelligence-cheating-australian-catholic-university/105863524 (2025-10-09)
• https://www.news.uct.ac.za/article/-2025-07-24-uct-scraps-flawed-ai-detectors (2025-07-24)
• https://campustechnology.com/articles/2024/05/08/turnitin-more-than-half-of-students-continue-to-use-ai-to-write-papers.aspx (2024-05-08)
• https://www.theregister.com/2025/06/16/university_ai_cheating/ (2025-06-16)
• https://www.timeshighereducation.com/news/student-ai-cheating-cases-soar-uk-universities (2024-11-19)
• https://openai.com/index/how-people-are-using-chatgpt/ (2025-07-01)
• https://sites.campbell.edu/academictechnology/2025/03/06/ai-in-higher-education-a-summary-of-recent-surveys-of-students-and-faculty/ (2025-03-06)
• https://programs.com/resources/students-using-ai/ (2025-01-01)
• https://online.ysu.edu/degrees/education/msed/ai-teaching-statistics-usage-and-trends/ (2025-01-01)
• https://newsroom.collegeboard.org/new-research-majority-high-school-students-use-generative-ai-schoolwork (2025-10-06)
• https://hai.stanford.edu/news/ai-detectors-biased-against-non-native-english-writers (2023-05-15)
• https://neurips.cc/virtual/2023/poster/71402 (2023-12-01)
• https://aclanthology.org/2026.eacl-srw.20.pdf (2026-01-01)
• https://www.researchsquare.com/article/rs-7359956/latest (2024-01-01)
• https://link.springer.com/article/10.1186/s41239-024-00487-w (2024-01-01)
• https://citl.news.niu.edu/2024/12/12/ai-detectors-an-ethical-minefield/ (2024-12-12)
• https://education.ufl.edu/eprc/files/2025/11/EPRC_PolicyBrief-AI-and-School-Discipline-Curran-and-Goo-2025.pdf (2025-10-01)
• https://www.usccr.gov/files/2024-12/2024-ai-in-education.pdf (2024-12-01)
• https://www.unesco.org/en/articles/unesco-survey-two-thirds-higher-education-institutions-have-or-are-developing-guidance-ai-use (2025-09-02)
• https://www.rand.org/pubs/research_reports/RRA956-31.html (2025-01-01)
• https://www.pewresearch.org/short-reads/2025/01/15/about-a-quarter-of-us-teens-have-used-chatgpt-for-schoolwork-double-the-share-in-2023/ (2025-01-15)
• https://www.abc.net.au/news/2025-10-20/universities-using-ai-to-detect-students-cheating/105905804 (2025-10-20)
• https://www.turnitin.com/blog/what-are-the-new-and-emerging-trends-in-academic-misconduct (2025-01-01)
• https://dgmg81phhvh63.cloudfront.net/content/user-photos/AACU_AI_Report_2025.pdf (2024-12-07)
• https://www.bcg.com/publications/2026/how-ai-can-help-universities-capture-opportunity (2026-01-01)
• https://www.ellucian.com/blog/ai-higher-education-2025-survey-findings-move-strategic-integration (2025-01-01)
• https://www.turnitin.com/blog/are-universities-ready-for-ai-native-academic-integrity (2025-02-26)
• https://www.pursuit.us/news/ai-in-education-news-policies-innovations (2026-03-05)